GA4 sessions(GA4)

Long definition

GA4 redefined sessions around event collection rather than the page-view-centric model of Universal Analytics. When the GA4 SDK detects activity after 30 minutes of inactivity, it fires a session_start event with a fresh ga_session_id. Every subsequent event in the session inherits that ID. The session count in reports is the number of distinct session_start events, not a separate metric.

Three behavioral differences from Universal Analytics deserve specific attention:

• Source/campaign changes do not split sessions. In UA, a user who arrived via organic search, then within the same window clicked a paid ad on another tab and returned, generated two sessions. In GA4 it is one session. Attribution still attaches the latest valid source, but session count is no longer inflated by mid-session source changes.
• The midnight cutoff is gone. UA terminated the session at midnight (account timezone). GA4 lets a session span midnight; date dimension is determined by when session_start fired.
• Engaged sessions are a separate metric. A "session" is the count of session starts; an "engaged session" is one that lasted longer than 10 seconds, fired a conversion event, or had at least 2 page views. Engaged sessions feed engagement rate; raw sessions feed traffic volume.

GA4 sessions are device-bound and cookie-bound. A user moving from mobile to desktop creates two sessions for two users (unless User-ID or Google Signals reconciliation is on). Cross-device deduplication happens in modeled identity reports, not in raw session counts.

For SEO reporting, the practical implications: session counts are slightly lower than equivalent UA setups because mid-session campaign changes no longer multiply them, and "Direct" sessions can be inflated when referrers are stripped. The GA4 sessions documentation covers the underlying logic.

Common misconceptions

• "GA4 sessions equal UA sessions, just renamed." They differ structurally. A side-by-side migration usually shows GA4 sessions running 5-15% lower than UA for the same traffic, driven primarily by the no-source-split rule and bot filtering improvements.
• "Session timeout is a tracking parameter you should tune." The 30-minute default rarely needs changing. Lengthening it inflates session duration metrics; shortening it inflates session counts. Both distort comparisons against industry benchmarks.
• "Sessions are the right primary KPI." For SEO, segment by landing page first, channel second, and look at engaged sessions or conversion-bearing sessions before reading raw session counts. Raw sessions include bot remnants, duplicates from referrer stripping, and bounce-equivalent traffic that matters to no one.
• "Direct traffic is real direct traffic." A meaningful share is dark-social or referrer-stripped organic. When Direct rises sharply with no campaign change, suspect a measurement issue (HTTPS-to-HTTP transition, app-to-web handoff) before celebrating brand strength.